Clinton Is Leading, and It Isn't Because of a Debate
General-election debates rarely cause major changes in voters’ choices. That’s what political scientists believe for the most part. Yet after the first presidential debate, on Sept. 26, Hillary Clinton moved back into a strong lead over Donald Trump. Indeed, according to the FiveThirtyEight forecasts, her chances of winning bottomed out on that very day (at 55 percent), and have increased steadily ever since.
So was the first debate a true game changer?
Polls fluctuate all the time in response to events that dominate the news. Sometimes, this is a result of real changes in the race. For example, once candidates clinch the presidential nomination, they typically pick up same-party supporters who backed one of their rivals in the primaries and caucuses. Then around the time of the conventions, the nominees add another wave of supporters: those who don’t pay much attention to politics and who use the events as signals to start tuning in and choosing a candidate.
But a lot of changes in the polls are predictably ephemeral. Something in the news that flatters one candidate makes marginal supporters more likely to say they have decided to vote for her or him and more likely to say they will vote, period. It even makes them more likely to answer a survey in the first place. Similarly, bad news about a candidate, if it dominates media reports, will make marginal supporters less likely to admit they have decided, less likely to claim they are certain to vote, and less likely to answer a survey.
All of that can be true -- producing noticeable changes in polling results -- even if nothing has really changed. And then when the good or bad news fades, polling returns to whatever the “normal” might be.
So most of what we’re seeing in long-term polling trendlines isn’t that a bunch of people are changing their minds back and forth. The underlying race can be fairly stable even while the polls fluctuate. That’s why analysts call these fluctuations “bounces” -- they go up and back down fairly predictably.
We don’t really know what “normal” is -- the equilibrium we would observe if the current news environment was relatively neutral.
What we can observe, however, is which news stories are getting plenty of play. And we know there was a run of bad news for Hillary Clinton in mid-September, centered on her comment that many Trump supporters were “deplorables” (reported on Sept. 10) and her bout with pneumonia (Sept. 11). Not only did those stories linger, but they also produced a polling dip, which was a third piece of bad news.
What happened in late September was the natural ebbing of that cycle, combined with a new round of negative stories about Trump. On the weekend before the first debate, several news articles detailed Trump’s problems with the truth. At the same time, a series of reports by the Washington Post’s David A. Fahrenthold and others about questionable practices involving Trump’s businesses and “charity” foundation entered the news cycle, replacing talk of Clinton’s health and “deplorables.” Several days after the debate, the New York Times ran an article on Trump’s 1990s tax returns.
The almost universal pans of Trump’s debate performance added to that avalanche of bad news. But the lasting story wasn’t the event itself; it was the story Clinton introduced at the debate about how Trump behaved toward a former Miss Universe. This issue remained in the news because Trump decided to fight a feud over it, and Clinton’s team had had an ad on the topic ready to roll out. Those developments were independent of the debate.
As I said, we don’t know where the polls would be in a period of relatively neutral news coverage. My guess is that Clinton is leading by something like 4 to 6 percentage points, and has throughout, with most polling surges in both directions short-term aberrations.
But that’s a judgment call; one certainly could argue that the underlying contest is somewhat tighter than this, and that we’ve had more negative cycles about Trump than about Clinton. You could also say that given a candidate with all of Trump’s liabilities, a neutral information environment would feature quite a few negative stories about him, so we need to build that into our expectations.
The bottom line: The first debate didn’t create Clinton’s current lead. So don’t expect the debates on Sunday or the one on Oct. 19 to be the reason if it changes.
I'm talking about polling averages. Individual polls, of course, will have additional sampling errors, and some have other problems, so looking at changes between individual polls will artificially enhance the appearance of instability. It's also possible that the polls as a whole could be biased for one or the other candidate, but that's a separate question.
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
To contact the author of this story:
Jonathan Bernstein at firstname.lastname@example.org
To contact the editor responsible for this story:
Katy Roberts at email@example.com